From: "Theodore Y. Ts'o" Subject: Re: =?utf-8?B?562U5aSNOiDnrZTlpI06IOetlA==?= =?utf-8?B?5aSNOiDnrZTlpI06IFtQQVRDSF0gZXh0NDogZTJmc3Byb2dzOiBmaXggaW5v?= =?utf-8?Q?de_bitma?= =?utf-8?Q?p?= num not integer,incompatible for ancient android devices Date: Sat, 30 Jun 2018 09:04:29 -0400 Message-ID: <20180630130429.GA26529@thunk.org> References: <1530014046-62466-1-git-send-email-gaoming20@huawei.com> <20180627140937.GA3348@thunk.org> <20180628022900.GA663@thunk.org> <20180628153022.GA8521@thunk.org> <20180629142612.GE1231@thunk.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: "linux-ext4@vger.kernel.org" , "linux-kernel@vger.kernel.org" , "Liqingchao (sorp)" , "Shenchen (harry)" , "miaoxie (A)" , "yangfei (D)" , "Renlipeng (OS driver)" To: "Gaoming (ming, consumer BG)" Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-ext4.vger.kernel.org On Sat, Jun 30, 2018 at 01:26:43AM +0000, Gaoming (ming, consumer BG) wrote: > Yes, it is caused by using 1024 blocksize. > It is historical problem, and I have to admit that's not good idea. I don't know why somebody choose it some years before. > It has been corrected two years before or more early. But some ancient devices exist. > It is not user data, no need to do file-based encryption. It is a small partition for some use. > > However, 1024 is legal though not good, somebody may use it. > And we should fix it. So you understand my position --- the reason why I've been pushing so hard is I'm trying to figure out how big of a problem this is. Specifically speaking, is this a Huawei-specific problem, or something across the entire Android ecosystem. I *thought* I had fixed most of the disaster back in 2011. There have periodic headaches where testers would discover problems where android handsets get bricked after doing a factory reset that I had tracked down to make_ext4fs, and the existence of make_ext4fs is not something I agreed to, and have been fighting for years. So I've been cleaning up after make_ext4fs for a while, even though it's not my responsiblity. (For one thing my work responsibilities are for data center servers at Google, *not* Android; for another, no one asked *me* before they came up with the abomination which is make_ext4fs.) So I don't feel particularly, or personally, responsible for bugs caused by make_ext4fs, because if it had been up to me, it would have never existed in the first place. If it's only in ancient Huawei devices, I don't see a strong reason to support his in upstream e2fsprogs. Are you really going to be backporting the latest e2fsprogs into these ancient Huawei devices? In my experience, the Android team has a hard enough time getting their Android partners to backport kernel fixes for severe security bugs into old Android devices --- never mind versions of e2fsprogs. If not, what's the point? Regards, - Ted